ObligationApplicablev1.0.0

Article 55: GPAI models with systemic risk

Additional duties for the most capable general-purpose AI models, on top of Article 53.

The official source remains authoritative. This is general information about obligations and not legal advice. See this object on the map

Address and citation

This object has an address of its own that is never renamed or reused. Store the identifier in your own file, not the title or the link.

Identifier: praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk
Payload hash (sha256): 112ee62609020ed4f49362f4910ee0bdb9155a472914dad7870a887141a5f589

Citation line

Praxikon, "Article 55: GPAI models with systemic risk", praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk@1.0.0, dataset praxikon:sys:registry:dataset:ai-act-implementation-graph 2.1.0 (schema 1.4.0), effective_at 2026-08-08T00:00:00.000Z, known_at 2026-08-14T00:00:00.000Z, sha256 112ee62609020ed4f49362f4910ee0bdb9155a472914dad7870a887141a5f589

Version: 1.0.0
Legal time (effective_at): 2 August 2025
Knowledge time (known_at): 8 August 2026
Closed on: Not closed
Application date: 2 August 2025
Topics: gpai-systemic-risk

Review status: Placed against the official source (8 August 2026). Next check due by 4 February 2027. The check date is the knowledge date of this version; no later recheck has been recorded.

Who must act and who is affected

On an obligation the addressee is recorded separately from the party that is merely affected. That difference decides who a supervisory authority addresses and who has work to do without the duty resting on them.

Duty holder

Provider of a GPAI modelpraxikon:eu:ai-act:actor:gpai-model-provider

Affected actor

Not recorded on this object.

Oversight

AI Officepraxikon:eu:ai-act:actor:ai-office

Evidence owner

Provider of a GPAI modelpraxikon:eu:ai-act:actor:gpai-model-provider

What this object links to

Every relation appears below as a path: from the source with its locator, through the conditions and exceptions of the object carrying the relation, to the consequence. A locator belongs to a statement in the data and not to a relation, so the source is the source anchor of the carrying object.

Who carries the duty

1 of 1 shown

The duty rests on this role. A supervisory authority addresses this role when it is not met.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- ActorProvider of a GPAI model
  praxikon:eu:ai-act:actor:gpai-model-provider

Who supervises this

1 of 1 shown

This role supervises and receives the notifications. It is the institutional role, not the national body that fills it.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- ActorAI Office
  praxikon:eu:ai-act:actor:ai-office

Who owns the evidence

1 of 1 shown

This duty holder creates, maintains and can produce the evidence.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- ActorProvider of a GPAI model
  praxikon:eu:ai-act:actor:gpai-model-provider

What has to be done

1 of 1 shown

This is what concretely has to be done under the obligation.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- ActionPerform model evaluations and risk mitigation
  praxikon:eu:ai-act:action:article-55-gpai-systemic-risk-act

What has to be recorded

1 of 1 shown

This is what has to be in the file to show that the action was carried out.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- EvidenceSystemic-risk file
  praxikon:eu:ai-act:evidence:article-55-gpai-systemic-risk-record

What keeps it in place

1 of 1 shown

This is what keeps compliance in place over time, periodic or event-driven.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- ControlCompute threshold monitoring
  praxikon:eu:ai-act:control:article-55-gpai-systemic-risk-control

Public tools

1 of 1 shown

This is the public route a reader can actually use.

Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
- Condition | allThe GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.
- ExceptionThe GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.
As long as this exception is not ruled out, the outcome stays conditional and you have to establish it yourself.
Consequence
- TemplateFull text of Article 55
  praxikon:eu:ai-act:template:article-55-gpai-systemic-risk-legal-text

What points at this object

Objects hanging off this obligation

7 of 7 shown

The object belongs to this obligation. The source line it hangs off sits there.

Source
This object carries no official fact of its own. The source line below sits on the obligation it hangs off; the link itself is recorded editorially.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: Perform model evaluations and risk mitigation
Consequence
- ActionPerform model evaluations and risk mitigation
  praxikon:eu:ai-act:action:article-55-gpai-systemic-risk-act
Source
Official fact on this object, with its locator.
- Guidelines for GPAI model providers
  Locator: Commission Guidelines C(2025) 5045 final, 18.7.2025
  praxikon:eu:ai-act:source:commission-gpai-guidelines
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: Guidelines on the scope of the GPAI obligations
Consequence
- ChangeGuidelines on the scope of the GPAI obligations
  praxikon:eu:ai-act:change:2025-07-18-gpai-guidelines
Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Articles 53 to 55 and Article 111(3)
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: GPAI model obligations apply
Consequence
- ChangeGPAI model obligations apply
  praxikon:eu:ai-act:change:2025-08-02-gpai-obligations-applicable
Source
Official fact on this object, with its locator.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 111(3)
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: Legacy GPAI models must comply
Consequence
- ChangeLegacy GPAI models must comply
  praxikon:eu:ai-act:change:2027-08-02-legacy-gpai-models-comply
Source
This object carries no official fact of its own. The source line below sits on the obligation it hangs off; the link itself is recorded editorially.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: Compute threshold monitoring
Consequence
- ControlCompute threshold monitoring
  praxikon:eu:ai-act:control:article-55-gpai-systemic-risk-control
Source
This object carries no official fact of its own. The source line below sits on the obligation it hangs off; the link itself is recorded editorially.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: Systemic-risk file
Consequence
- EvidenceSystemic-risk file
  praxikon:eu:ai-act:evidence:article-55-gpai-systemic-risk-record
Source
This object carries no official fact of its own. The source line below sits on the obligation it hangs off; the link itself is recorded editorially.
- EU Artificial Intelligence Act 2024/1689
  Locator: Article 55(1)-(2) with Article 51 and Article 52
  praxikon:eu:ai-act:source:reg-eu-2024-1689
  Open official source
Via
No condition or exception recorded on this object.
Relation recorded on: Full text of Article 55
Consequence
- TemplateFull text of Article 55
  praxikon:eu:ai-act:template:article-55-gpai-systemic-risk-legal-text

What this object states

Official fact

Attributable to a named primary source, with a locator. Where they differ, the official source prevails.

Article 55 obliges providers of GPAI models with systemic risk to perform model evaluations including adversarial testing, assess and mitigate systemic risks at Union level, track and report serious incidents, and ensure adequate cybersecurity of model and infrastructure. The duties apply since 2 August 2025; AI Office enforcement is active since 2 August 2026.
- Locator: Article 55(1)-(2) with Article 51 and Article 52praxikon:eu:ai-act:source:reg-eu-2024-1689Open official source

Our interpretation

Our own reading. It can change without the law changing, and it is not the position of a supervisory authority.

The 10^25 FLOPs threshold is a presumption, not a shield: the Commission can also designate models on capabilities, and fine-tuning on top of an existing model can in some circumstances trigger a qualification of its own.
- Locator: Article 55(1)-(2) with Article 51 and Article 52praxikon:eu:ai-act:source:reg-eu-2024-1689Open official source

Recommended step

A practical step we consider appropriate. Not an obligation following from the Regulation.

Model providers near the threshold: set up compute monitoring now and join the Code of Practice to carry the burden of proof.
- Locator: Article 55(1)-(2) with Article 51 and Article 52praxikon:eu:ai-act:source:reg-eu-2024-1689Open official source

When this applies

1The GPAI model has high-impact capabilities, presumed above 10^25 FLOPs of cumulative training compute, or is designated by the Commission.

When this does not apply

The GPAI Code of Practice can, following the adequacy assessment, serve as a means to demonstrate compliance.

Referring to this object

Citation block

Copy this reference into your advice, article or file. The identifier, the version and the hash keep the statement findable later, even once the dataset has moved on.

Reference

Praxikon, "Article 55: GPAI models with systemic risk",
praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk@1.0.0,
dataset praxikon:sys:registry:dataset:ai-act-implementation-graph 2.1.0 (schema 1.4.0),
effective_at 2025-08-02T00:00:00.000Z, known_at 2026-08-08T00:00:00.000Z,
sha256 112ee62609020ed4f49362f4910ee0bdb9155a472914dad7870a887141a5f589,
https://www.praxikon.com/en/verkenner/obligation/article-55-gpai-systemic-risk
(https://www.praxikon.com/api/v1/obligations?id=praxikon%3Aeu%3Aai-act%3Aobligation%3Aarticle-55-gpai-systemic-risk&effective_at=2025-08-02&known_at=2026-08-08&lang=en, accessed 2026-08-25)

Short form

praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk@1.0.0 (sha256 112ee626)

BibTeX

@misc{praxikon-eu-ai-act-obligation-article-55-gpai-systemic-risk-1-0-0,
  author       = {{Praxikon}},
  title        = {Article 55: GPAI models with systemic risk},
  year         = {2026},
  version      = {1.0.0},
  number       = {praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk},
  howpublished = {AI Act Change \& Evidence Graph, dataset 2.1.0, schema 1.4.0},
  note         = {effective_at 2025-08-02T00:00:00.000Z; known_at 2026-08-08T00:00:00.000Z; sha256 112ee62609020ed4f49362f4910ee0bdb9155a472914dad7870a887141a5f589},
  url          = {https://www.praxikon.com/en/verkenner/obligation/article-55-gpai-systemic-risk},
  urldate      = {2026-08-25},
  language     = {en}
}

CSL JSON

[
  {
    "id": "praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk@1.0.0",
    "type": "dataset",
    "title": "Article 55: GPAI models with systemic risk",
    "container-title": "AI Act Change & Evidence Graph",
    "publisher": "Praxikon",
    "version": "1.0.0",
    "number": "praxikon:eu:ai-act:obligation:article-55-gpai-systemic-risk",
    "URL": "https://www.praxikon.com/en/verkenner/obligation/article-55-gpai-systemic-risk",
    "language": "en",
    "issued": {
      "date-parts": [
        [
          2026,
          8,
          8
        ]
      ]
    },
    "accessed": {
      "date-parts": [
        [
          2026,
          8,
          25
        ]
      ]
    },
    "note": "dataset praxikon:sys:registry:dataset:ai-act-implementation-graph 2.1.0; schema 1.4.0; effective_at 2025-08-02T00:00:00.000Z; known_at 2026-08-08T00:00:00.000Z; sha256 112ee62609020ed4f49362f4910ee0bdb9155a472914dad7870a887141a5f589; retrieved_from https://www.praxikon.com/api/v1/obligations?id=praxikon%3Aeu%3Aai-act%3Aobligation%3Aarticle-55-gpai-systemic-risk&effective_at=2025-08-02&known_at=2026-08-08&lang=en; licence https://www.praxikon.com/nl/legal/terms"
  }
]

How to verify a reference later is set out in the methodology. Terms

For agents and integrations

This page and the machine output come from the same object and the same two time axes.

Open as JSON Open as JSON-LD Human explanation Read the methodology